A Survey of Automatic Sleep Staging: Technical Route, Technical Challenges, and Future Perspectives

A complete human sleep consists of multiple states, each of which has its own unique and typical characteristics. A clear definition of these states lays a solid foundation for the study of related sleep disorders. However, there remain significant challenges in current research on human sleep. Focusing on the automatic staging of sleep states, this review is organized around the technical route, technical challenges, and future perspectives of sleep staging. First, 2044 records related to sleep staging were initially screened from the Web of Science Core Collection, and key methodological themes were identified through an initial screening. Second, more than 200 highly relevant studies were selected for review, and 174 of them were cited. Compared with previous surveys on automated sleep staging, in this paper, more attention is paid to the complete technical route, key technical challenges, data strategy, and future research trends. Special attention is given to the N1 stage, which represents the most challenging classification target due to both its inherent physiological ambiguity and low inter-scorer reliability under the current AASM framework. This review aims to provide a practically oriented and operationally actionable reference for researchers in the field.